Improved Retinex and Multi-Image Fusion Algorithm for Low Illumination Image Enhancemen
-
摘要: 为了解决低照度图像在图像增强过程中图像质量不佳、对比度不高等问题,本文提出改进Retinex与多图像融合算法用于低照度图像增强。首先将待处理图像转换到HSV色彩空间,并设定阈值对其V通道分量进行亮度调节,然后转换到RGB色彩空间,将其拷贝3份,对第一份进行直方图均衡化,中值滤波处理;对第2份进行自动亮度调节,双边滤波处理;对第3份进行改进的Retinex算法处理,采用高斯滤波、双边滤波作为其环绕函数,估计图像照明分量,最后输出反射图。将处理后的3份图像转到HSV色彩空间,对其V分量进行多图像融合,H、S分量沿用第2份图像分量值,最后将融合后的图像由HSV转为RGB色彩空间,输出处理后的图像。实验结果表明,本文提出的算法在增强低照度图像的同时,还可抑制图像噪声,同时具有良好的保边性,且细节明显。Abstract: To solve the problems of poor image quality and low contrast in low-illumination image enhancement, this study proposes an improved Retinex and multi-image fusion algorithm for low -illumination image enhancement. First, the image to be processed is converted to the HSV color space, and the brightness of the V-channel component is adjusted by setting a threshold. Then, it is converted to the RGB color space, and three copies are made. Histogram equalization and median filtering are performed for the first part; the second part is processed by automatic brightness adjustment and bilateral filtering; the third part is processed by an improved Retinex algorithm, which uses Gaussian filtering and bilateral filtering as its surround function to estimate the illumination component of the image, and outputs the reflection image. The three processed images are transferred to the HSV color space, and the V component is fused. The H and S components follow the values of the second image component. Finally, the fused image is converted from the HSV to RGB color space, and the processed image is output. The experimental results show that the proposed algorithm not only enhances the low-illumination image but also suppresses the image noise. Furthermore, it exhibits good edge preservation and obvious details.
-
Key words:
- low illumination /
- image enhancement /
- Retinex /
- bilateral filtering /
- image component fusion /
- HSV color space
-
表 1 Label 1图像增强质量评价
Table 1. Label 1 image enhancement quality evaluation
Label 1 Label diagram SSR MSR Gamma Fu’s algorithm Ours PSNR 14.51 4.43 4.89 17.32 18.20 19.45 SSIM 0.58 0.35 0.40 0.82 0.87 0.92 IE 7.15 5.54 6.46 5.43 7.12 6.90 表 2 Label 5图像增强质量评价
Table 2. Label 5 image enhancement quality evaluation
Label 5 Label diagram SSR MSR Gamma Fu’s algorithm算法 Ours PSNR 13.41 3.70 4.48 14.00 15.41 16.56 SSIM 0.44 0.34 0.32 0.22 0.59 0.70 IE 7.2 6.29 6.62 5.37 6.34 7.04 -
[1] 闫保中, 韩旭东, 何伟. 基于Retinex理论改进的低照度图像增强算法[J]. 应用科技, 2020, 47(5): 74-78. https://www.cnki.com.cn/Article/CJFDTOTAL-YYKJ202005013.htmYAN Baozhong, HAN Xudong, HE Wei. Improved low illumination image enhancement algorithm based on Retinex theory[J]. Applied Science and Technology, 2020, 47(5): 74-78. https://www.cnki.com.cn/Article/CJFDTOTAL-YYKJ202005013.htm [2] 牟琦, 魏妍妍, 李姣, 等. 改进的Retinex低照度图像增强算法研究[J]. 哈尔滨工程大学学报, 2018, 39(12): 2001-2010. https://www.cnki.com.cn/Article/CJFDTOTAL-HEBG201812019.htmMOU Qi, WEI Yanyan, LI Jiao, et al. Research on improved Retinex low illumination image enhancement algorithm[J]. Journal of Harbin Engineering University, 2018, 39(12): 2001-2010. https://www.cnki.com.cn/Article/CJFDTOTAL-HEBG201812019.htm [3] 韩梦妍, 李良荣, 蒋凯. 基于光照图估计的Retinex低照度图像增强算法研究[J/OL]. [2020-07-28]. 计算机工程, http://www.ecice06.com/CN/10.19678/j.issn.1000-3428.0059224.HAN Mengyan, LI Liangrong, JIANG Kai. Research on Retinex low illumination image enhancement algorithm based on illumination map estimation[J/OL]. [2020-07-28]. Computer Engineering, http://www.ecice06.com/CN/10.19678/j.issn.1000-3428.0059224. [4] 赵馨宇, 黄福珍. 基于双通道先验和光照图引导滤波的图像增强[J]. 激光与光电子学进展, 2021, 58(8): 53-62. https://www.cnki.com.cn/Article/CJFDTOTAL-JGDJ202108005.htmZHAO Xinyu, HUANG Fuzhen. Image enhancement based on dual channel a priori and illumination guided filtering[J]. Progress in Laser and Optoelectronics, 2021, 58(8): 53-62. https://www.cnki.com.cn/Article/CJFDTOTAL-JGDJ202108005.htm [5] 刘明珠, 苏桃宇, 王依宁. 压扩多尺度Retinex图像增强算法研究[J]. 哈尔滨理工大学学报, 2020, 25(5): 93-99. https://www.cnki.com.cn/Article/CJFDTOTAL-HLGX202005014.htmLIU Mingzhu, SU Taoyu, WANG Yining. Research on companding multiscale Retinex image enhancement algorithm[J]. Journal of Harbin University of Technology, 2020, 25(5): 93-99. https://www.cnki.com.cn/Article/CJFDTOTAL-HLGX202005014.htm [6] SUN J, HE K, TANG X. Single Image Haze Removal Using Dark Channel Priors: U.S. Patent 8, 340, 461[P]. 2012-12-25. [7] CHEN C, CHEN Q, XU J, et al. Learning to See in the Dark [C/OL][2018-05-04]//Conference on Computer Vision and Pattern Recognition(CVPR), https://arxiv.org/abs/1805.01934. [8] 李斯娜, 明道洋. 小波变换在图像去噪中的应用[J]. 科技信息, 2013(18): 253. https://www.cnki.com.cn/Article/CJFDTOTAL-KJXX201318215.htmLI SinA, MING Daoyang. Application of wavelet transform in image denoising [J]. Science and Technology Information, 2013(18): 253. https://www.cnki.com.cn/Article/CJFDTOTAL-KJXX201318215.htm [9] FU X Y, ZENG D L, HUANG Y, et al. A fusion-based enhancing method for weakly illuminated images[J]. Signal Processing, 2016, 129: 82-96. doi: 10.1016/j.sigpro.2016.05.031 [10] 秦绪佳, 王慧玲, 杜轶诚, 等. HSV色彩空间的Retinex结构光图像增强算法[J]. 计算机辅助设计与图形学学报, 2013(4): 488-493. doi: 10.3969/j.issn.1003-9775.2013.04.008QIN Xujia, WANG Huiling, DU Yicheng, et al. Retinex structured light image enhancement algorithm in HSV color space[J]. Journal of Computer Aided Design and Graphics, 2013(4): 488-493. doi: 10.3969/j.issn.1003-9775.2013.04.008 [11] 杨振亚, 王勇, 杨振东, 等. RGB颜色空间的矢量-角度距离色差公式[J]. 计算机工程与应用, 2010(6): 154-156. https://www.cnki.com.cn/Article/CJFDTOTAL-JSGG201006045.htmYANG Zhenya, WANG Yong, YANG Zhendong, et al. Vector angle distance color difference formula of RGB color space[J]. Computer Engineering and Application, 2010(6): 154-156. https://www.cnki.com.cn/Article/CJFDTOTAL-JSGG201006045.htm [12] 武治国, 王延杰. 一种基于直方图非线性变换的图像对比度增强方法[J]. 光子学报, 2010, 39(4): 755-758. https://www.cnki.com.cn/Article/CJFDTOTAL-GZXB201004037.htmWU Zhiguo, WANG Yanjie. An image contrast enhancement method based on histogram nonlinear transformation[J]. Acta Photonica Sinica, 2010, 39(4): 755-758. https://www.cnki.com.cn/Article/CJFDTOTAL-GZXB201004037.htm [13] 孙抗, 汪渤, 周志强, 等. 基于双边滤波的实时图像去雾技术研究[J]. 北京理工大学学报, 2011(7): 810-813. https://www.cnki.com.cn/Article/CJFDTOTAL-BJLG201107015.htmSUN Kang, WANG Bo, ZHOU Zhiqiang, et al. Research on real-time image defogging technology based on bilateral filtering[J]. Journal of Beijing University of Technology, 2011(7): 810-813. https://www.cnki.com.cn/Article/CJFDTOTAL-BJLG201107015.htm [14] 张亚飞. 基于幂次变换和MSR的光照不均图像增强[J]. 电脑知识与技术, 2012, 8(22): 5456-5458. https://www.cnki.com.cn/Article/CJFDTOTAL-DNZS201222067.htmZHANG Yafei. Uneven illumination image enhancement based on power transform and MSR[J]. Computer Knowledge and Technology, 2012, 8 (22): 5456-5458. https://www.cnki.com.cn/Article/CJFDTOTAL-DNZS201222067.htm [15] 杨先凤, 李小兰, 贵红军. 改进的自适应伽马变换图像增强算法仿真[J]. 计算机仿真, 2020, 37(5): 246-250. https://www.cnki.com.cn/Article/CJFDTOTAL-JSJZ202005048.htmYANG Xianfeng, LI Xiaolan, GUI Hongjun. Simulation of improved adaptive gamma transform image enhancement algorithm[J]. Computer Simulation, 2020, 37 (5): 246-250. https://www.cnki.com.cn/Article/CJFDTOTAL-JSJZ202005048.htm [16] 郭璠, 蔡自兴. 图像去雾算法清晰化效果客观评价方法[J]. 自动化学报, 2012, 38(9): 1410-1419. https://www.cnki.com.cn/Article/CJFDTOTAL-MOTO201209004.htmGUO Xuan, CAI Zixing. Objective evaluation method of image defogging algorithm clarity effect[J]. Journal of Automation, 2012, 38 (9): 1410-1419. https://www.cnki.com.cn/Article/CJFDTOTAL-MOTO201209004.htm